# Phase 1 Implementation Walkthrough

The foundational Phase 1 (Retrieval-Only Architecture) for your Unified Email Semantic Search system is now complete. The repository structure, Docker configurations, backend API, indexing daemon, and frontend have been successfully scaffolded in `/home/adipu/AllMail`.

## Repository Structure Overview

```mermaid
graph TD
    A[AllMail/] --> B(db_data/)
    A --> C(Maildir/)
    A --> D(mail-sync/)
    A --> E(api/)
    A --> F(webmail/)
    A --> G(.gitea/)
    A --> H(docker-compose.yml)
    A --> I(.env.example)

    D --> D1[Dockerfile]
    D --> D2[entrypoint.sh]
    D --> D3[mbsyncrc.template]

    E --> E1[Dockerfile]
    E --> E2[main.py]
    E --> E3[indexer.py]
    E --> E4[database.py]
    E --> E5[run.sh]

    F --> F1[Dockerfile]
    F --> F2[nginx.conf]
    F --> F3[entrypoint.sh]
    F --> F4[spa/]
```

## Key Components Implemented

### 1. Docker Infrastructure & Resource Management
- Created `docker-compose.yml` with memory limits applied specifically to respect the 16GB total memory overhead:
  - `db` is restricted to `1G` (and Postgres tuned via command arguments `shared_buffers=256MB`).
  - `api` is restricted to `512M`.
  - `mail-sync` is restricted to `256M`.
  - `webmail` is restricted to `512M`.
- Uses real filesystem mounts for persistent data: `./db_data` for PostgreSQL and `./Maildir` for the email files.

### 2. Database & API (`api/`)
- A Python 3.11 container running both a FastAPI server (`main.py`) and a background Watchdog daemon (`indexer.py`).
- Integrates `pgvector` and uses SQLAlchemy to manage vector embeddings.
- Automatically connects to OpenAI for `text-embedding-3-small` and indexes new files arriving in `./Maildir`.

### 3. Synchronization (`mail-sync/`)
- A lightweight Alpine image utilizing `isync` (mbsync).
- Contains an entrypoint that initializes a 5-minute cron schedule.
- A `mbsyncrc.template` was created to serve as your configuration starting point.

### 4. Frontend & SnappyMail (`webmail/`)
- A custom `php:8.2-fpm-alpine` container running Nginx.
- Serves a React/Vite SPA on the root (`/`) for querying the Unified Search API.
- Proxies `/mail/` to SnappyMail for standard webmail consumption.
- The `Dockerfile` handles building the React app automatically during `docker compose build`.

### 5. Deployment CI/CD (`.gitea/`)
- A GitHub-Actions-compatible workflow is present at `.gitea/workflows/deploy.yml` which triggers on pushes to `main` and runs the requested commands (`docker compose up -d --build` & `docker image prune -f`).

## Next Steps for You

> [!IMPORTANT]
> **Action Required**: The system is ready to be started, but you need to configure your secrets and accounts before running `docker compose up -d`.

1. **Configure Environment Variables**:
   Copy `.env.example` to `.env` and fill in your passwords and the OpenAI API Key.
   ```bash
   cp /home/adipu/AllMail/.env.example /home/adipu/AllMail/.env
   # Edit .env
   ```

2. **Configure Mail Accounts**:
   Copy the `mbsyncrc` template to an actual config file and configure your IMAP servers.
   ```bash
   cp /home/adipu/AllMail/mail-sync/mbsyncrc.template /home/adipu/AllMail/mail-sync/mbsyncrc
   # Edit mbsyncrc
   ```

3. **Deploy Phase 1**:
   After configuring the above files, you can start the system locally to ensure it builds correctly:
   ```bash
   cd /home/adipu/AllMail
   docker compose up -d --build
   ```